feat(citability): logischer norm_id-Join auf legal_basis (KB-v2 Zitier-Vertrag)

Wake-up #2 (Domaene 2): Zitierfaehigkeit ohne char-Level-Spans via logischem norm_id-Join auf KB-v2-Units (bp_compliance_kb_2026_1_build). Konvention (Board Compliance/KB-v2 2026-07-01): EU-<ACT>-Anhang<ROM> (Annex-Ebene, confirmed) / EU-<ACT>-Art<N> + EU-<ACT>-Kapitel<ROM> (verify_pending). Namensvariante EU-MaschVO-* (NICHT MaschinenVO). KEINE neue Klasse — norm_ids ist ein Attribut auf legal_basis (freeze-safe). - 65/65 legal_basis gejoint (CRA 40 + MaschVO 25), 0 unparsed; 64 Obligations citation_status -> norm_id_linked (BP/guidance-anchored bleiben ohne norm_id). - 53 annex_confirmed, 12 verify_pending; distinkt 5 Annex-IDs + 19 Art/Kapitel. - norm_id_manifest.json = KB-v2-Handoff (verify_pending Art-/Kapitel-IDs pruefen). - Granularitaet annex-grob (Part/Punkt = KB-Enhancement TBD); Artikel-norm_ids in KB-v2 noch zu verifizieren. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
2026-07-01 12:14:55 +02:00
parent c7339e68df
commit ffbedfa0dc
10 changed files with 692 additions and 171 deletions
@@ -0,0 +1,75 @@
+"""Zitierfähigkeits-Join (Wake-up #2): logischer norm_id-Join auf legal_basis.
+
+KB-v2-Konvention (Board 2026-07-01, Compliance/KB-v2): `EU-<ACT>-Anhang<ROM>` (Annex-Ebene, grob) ·
+`EU-<ACT>-Art<N>` (Artikel, in KB-v2 noch zu verifizieren) · Kapitel = TBD-Konvention.
+Namensvariante: `EU-MaschVO-*` (NICHT MaschinenVO). Kein char-Span nötig — logischer Join auf norm_id.
+Fügt `norm_ids` (Liste) je legal_basis + `norm_id_status` hinzu; setzt obligation.citation_status
+auf `norm_id_linked` (annex-grob). KEINE neue Klasse (Attribut). Freeze-safe.
+"""
+from __future__ import annotations
+
+import glob
+import json
+import re
+
+ACT = {"CRA": "CRA", "MaschVO": "MaschVO", "MaschinenVO": "MaschVO"}
+FILES = sorted(glob.glob("obligations/cra*.json"))
+
+
+def derive(source: str, anchor: str) -> tuple[list[str], str]:
+    act = ACT.get(source, source)
+    ids: list[str] = []
+    for rom in re.findall(r"An(?:hang|nex)\s+([IVX]+)", anchor, re.I):
+        ids.append(f"EU-{act}-Anhang{rom.upper()}")
+    articles = re.findall(r"\bArt(?:icle|\.)?\s*(\d+)", anchor)
+    chapters = re.findall(r"Kapitel\s+([IVX/]+)", anchor, re.I)
+    verify: list[str] = []
+    for n in articles:
+        verify.append(f"EU-{act}-Art{n}")
+    for grp in chapters:
+        for rom in grp.split("/"):
+            rom = rom.strip()
+            if rom:
+                verify.append(f"EU-{act}-Kapitel{rom.upper()}")
+    # dedup, Annexe zuerst (confirmed), dann verify
+    seen: set[str] = set()
+    ordered = [x for x in ids + verify if not (x in seen or seen.add(x))]
+    status = "annex_confirmed" if ids else ("verify_pending" if verify else "unparsed")
+    return ordered, status
+
+
+def main() -> None:
+    total_lb = linked = unparsed = 0
+    obl_linked = 0
+    for f in FILES:
+        d = json.load(open(f, encoding="utf-8"))
+        d.setdefault("norm_id_contract", {
+            "convention": "EU-<ACT>-Anhang<ROM> (Annex-Ebene) / EU-<ACT>-Art<N> (verify) — KB-v2 bp_compliance_kb_2026_1_build",
+            "act_naming": "EU-MaschVO-* (NICHT MaschinenVO)",
+            "granularity": "annex-grob — 'Annex I Part II (1)' -> EU-CRA-AnhangI; Part/Punkt = KB-Enhancement TBD",
+            "article_status": "EU-<ACT>-Art<N> in KB-v2 noch zu verifizieren; Annex-IDs confirmed",
+            "source": "Board Compliance/KB-v2 2026-07-01",
+        })
+        for o in d.get("obligations", []):
+            got = False
+            for b in o.get("legal_basis", []):
+                total_lb += 1
+                nids, st = derive(b.get("source", ""), b.get("anchor", ""))
+                b["norm_ids"] = nids
+                b["norm_id_status"] = st
+                if nids:
+                    linked += 1
+                    got = True
+                if st == "unparsed":
+                    unparsed += 1
+                    print(f"  UNPARSED: {b.get('source')} \"{b.get('anchor')}\"")
+            if got:
+                o["citation_status"] = "norm_id_linked"
+                obl_linked += 1
+        json.dump(d, open(f, "w", encoding="utf-8"), ensure_ascii=False, indent=1)
+    print(f"legal_basis gesamt {total_lb} | mit norm_ids {linked} | unparsed {unparsed}")
+    print(f"Obligations citation_status -> norm_id_linked: {obl_linked}")
+
+
+if __name__ == "__main__":
+    main()